A temperature sensor can be used to monitor temperature of an electronic component such as a CPU (Central Processing Unit), CPU (Graphics Processing Unit), MPU (Micro Processing Unit), SOC (System On Chip), etc. When the temperature exceeds predetermined thresholds, the sensor may alert a circuitry to slow down or even shut down the unit to reduce power consumption and thus reduce the temperature so that overheat that can cause destructive failure to the unit may be prevented.
Typically, temperature sensors include a reference circuitry and a temperature measuring circuitry wherein the temperature dependency is either proportional to absolute temperature (PTAT) or complementary to absolute temperature (CTAT). Further, DAC (digital to analog converter) based temperature sensors relying on comparing a PTAT voltage and a CTAT base-emitter voltage may be used. This approach, however, suffers from DAC code-to-temperature non-linearity issues, i.e., it cannot achieve good linearity over a wide temperature range, resulting in poor temperature measurement accuracy.
In some approaches involving a CTAT voltage, the compared voltage varies with high temperature coefficient over the design temperature range. When the compared voltage is PTAT, the implementation scheme introduces an intrinsic DAC code-to-temperature nonlinearity, and, as a result, suffers from poor temperature measurement accuracy unless an extensive temperature calibration (e.g., many-point calibration) is performed. Other approaches trying to cause the compared (or reference) voltage family curves to be parallel, and thus better DAC code-to-temperature linearity, are far from successful because the curves, in reality, are not parallel. These approaches therefore also suffer from poor temperature measurement accuracy.